Pharmacogenetics Research Network and Knowledge Base Third Scientiﬁc Meeting Alison F

Meeting Report 437

Pharmacogenetics Research Network and Knowledge Base Third Scientiﬁc Meeting Alison F. Davisa and Rochelle M. Longb

Pharmacogenetics 2003, 13:437–440

Keywords: PharmGKB, pharmacogenetics, pharmacogenomics, National Institutes of Health, population genetics

aOfﬁce of Communications and Public Liaison and bDivision of Pharmacology, Physiology and Biological Chemistry, National Institute of General Medical Sciences, National Institutes of Health, Bethesda, Maryland, USA.

Correspondence and requests for reprints to Rochelle M. Long, Division of Pharmacology, Physiology and Biological Chemistry, 45 Center Dr., Rm. 2AS-49G, National Institute of General Medical Sciences, National Institutes of Health, Bethesda, MD 20892-6200, USA. Tel: +301 594 1826; fax: +301 480 2802; e-mail: [email protected]

Received 2 April 2003 Accepted 10 May 2003

Introduction Quantitative and analytical challenges in The Pharmacogenetics Research Network and Know- genetics research ledge Base (PGRN) held its third annual open scientific While the history of pharmacogenetics research dates meeting on 3 March 2003 in Memphis, Tennessee. back nearly 50 years, changes in the scientific landscape The PGRN effort is a nationwide pharmacogenetics have refocused attention on current challenges facing research initiative sponsored by the National Institutes this field. Investigating multigenic human variation, of Health. Network members submit data to a compre especially as it relates to the molecular basis of complex hensive knowledge base, PharmGKB (http://www. diseases, is a critical element of modern genetics pharmGKB.org/). The meeting focused on the analyti research. While Mendelian disease genes are typically cal and quantitative issues confronting modern genetics of recent origin and alleles have large phenotypic research. Morning sessions addressed population genet effects, complex disease traits are often characterized ics, analytical challenges in molecular epidemiology by alleles exerting small phenotypic effects. In studying and comparative genomics. Afternoon sessions featured complex diseases, Aravinda Chakravarti of The Johns selected research updates from PGRN investigators. Hopkins University School of Medicine noted that functional tests of alleles and gene interactions must The PGRN was conceived as a network of multidisci complement gene mapping efforts, and that an essen plinary, collaborative research groups of investigators tial component of these studies is recapitulation of the examining how genetic variation contributes to inter- disease trait in an experimental system. Good study individual differences in drug responses. The groups design can tease apart the many factors that produce a are charged with collecting comprehensive, integrative complex disease, including episodic and age-dependent information about proteins and gene families known to traits; gender, environment and lifestyle; unknown be important in mediating the effects of therapeutics, gene actions; positive or negative regulatory alleles; and from variations in drug metabolizing enzymes to target epigenetic phenomena. receptors. Once comprehensive biological information for particular proteins and genes and their correspond Chakravarti described recent data from his group on ing variants is known, the network will catalogue this the complex genetics of Hirschsprung disease, in which information in a manner that is accessible and inter pedigree studies of Mennonite populations combined pretable to a wide range of scientists. PharmGKB has with a genome-wide association screen have recently the capacity to collect genetic sequence information, revealed two haplotypes in the receptor tyrosine kinase display polymorphic forms, identify functional conse (RET) gene [2]. Although variant alleles of several quences of genetic variation and rigorously correlate genes have been associated with this condition, none is this information with clinical drug responses [1]. necessary or sufficient to cause Hirschsprung disease.

0960-314X & 2003 Lippincott Williams & Wilkins DOI: 10.1097/01.fpc.0000054101.48725.49 Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 438 Pharmacogenetics 2003, Vol 13 No 7

Chakravarti reported that linkage disequilibrium stud about variation in drug metabolizing enzymes, as well ies have uncovered a genetic interaction between the as to impact the design and interpretation of preclinical endothelin receptor B (EDNRB) gene and multiple investigations. Anna Di Rienzo of the University of RET loci, and that preferential transmission of haplo Chicago discussed comparative genomics approaches to types may contribute to the phenotype of this disease. resequencing studies, in which the same sequence block is analysed in multiple individuals. Di Rienzo Michael Province of Washington University in St Louis described ongoing efforts of the Pharmacogenetics of discussed means to controlling type I error in statistical Anticancer Agents Research Group to clarify the genet data analysis. Province described the trade-offs inherent ic basis for toxicity in response to irinotecan and to selecting genotyping strategies, noting that research flavopiridol therapy. Comparative genomics analyses in ers pay a statistical penalty for excessive genotyping. the human, dog, rat, mouse and primate versions of the More data do not necessarily imply more information UGT1A gene cluster have begun to identify conserved and, for any complex trait, most genetic variation is regions, using the MultiPipMaker computational tool noise (including that present in candidate genes). To (http://bio.cse.psu.edu/pipmaker) [6]. This Web-based help explain the phenomenon, Province presented an tool draws percentage identity plots (pips) by compar analogy describing catching tuna with a fishing net, ing several genomic sequences, then creating align through which an occasional dolphin passes. In this ments and identifying nucleotide identities. One of the case, the tuna represent the test hypothesis, and the technique’s strengths is its unbiased analysis of any dolphins are the null hypothesis. In the example, larger DNA sequence, regardless of whether the sequence is and more prevalent passage of dolphins constitutes coding or non-coding. Di Rienzo predicted that future ‘noise’; however, increasing the fineness of the net (e.g. enhancements to comparative genomics strategies by applying a Bonferroni correction) severely restricts would include methods to assess statistical significance, data collection such that hardly any tuna can swim large-scale studies to measure functional significance through the net. This type of statistical correction can and new techniques to take into account phylogenetic produce a large power loss, restricting the utility of distances between compared species. such an approach. PGRN research update Province suggested that pharmacogenetics research Selected members of the PGRN groups presented demands a separation between gene discovery and recent research findings on adrenergic pharmacology, hypothesis testing. An approach pioneered in the 1940s membrane transporters and the pharmacogenomics of by Hungarian mathematician Abraham Wald led to the chemotherapy and asthma treatment. development of a current technique called sequential multiple decision procedures, which can help remedy Stephen Liggett of the University of Cincinnati pre the problem of balancing false positives (type I error) sented recent findings on a synergistic interaction be- and false negatives (type II error). Using this method tween 1 and Æ2C-adrenergic receptor polymorphisms lets the experimental data determine the point at which that may contribute to the pathology of congestive a hypothesis-generation phase should move to a true heart failure. G protein-coupled receptors constitute hypothesis-testing phase [3,4]. The approach helps to the largest family of signalling molecules, many of optimize sample size, while limiting both false positives which are involved in conveying drug action. Substan and false negatives. tial interindividual variation exists in agonist and antagonist response directed to these receptors; the Joseph Terwilliger of Columbia University contrasted known variable response to -blocker therapy repre the goals and approaches of genetics and epidemiology. sents a particularly relevant topic for pharmacogenetics Whereas epidemiology is hypothesis-based and an investigation. Some of this variation may be attributed ascertainment bias is undesirable, genome-scanning to polymorphisms in coding regions, which may act experiments are hypothesis-free and an ascertainment together to provide persistent adrenergic stimulation bias is necessary to maximize the efficiency of mapping and promote heart failure and/or cardiac hypertrophy. studies. Genetic studies with small sample sizes tend to Liggett described a small clinical study in which control systematically overestimate linkage disequilibrium and or heart failure patients were genotyped at both the 1 underestimate variation [5]. Terwilliger expressed the and Æ2C-adrenergic receptor loci. Logistic regression view that genetic approaches need an ascertainment analysis was used to determine the potential effect of bias because penetrance is not equivalent to detection, each genotype and the interaction between them on and that genome-scanning methods should be limited the risk of heart failure. Data acquired from the African to generating hypotheses, not to testing them. American cohort in this study were analysed separately, revealing a significantly increased odds ratio for risk of Pharmacogenetics research may benefit from using heart failure in African-American subjects who are comparative genomics strategies to glean information homozygous for both polymorphisms [7].

Kathleen Giacomini, Ira Herskowitz, and colleagues at number), as well as variable gene expression, gene the University of California, San Francisco (UCSF) are fusions and gene translocations. Additional complexity in the process of cataloguing the extent of human arises from the fact that adult and childhood cancer sequence variation in membrane transporters, which are differ at the molecular level. Evans reported the impacted by many classes of medications. The group is molecular classification of ALL based on gene expres analysing human DNA for the existence of single sion profiling [11]. The primary microarray data have nucleotide polymorphisms (SNPs) in the ATP-binding been deposited in the pharmacogenetics knowledge cassette transporter superfamily and in solute carrier base, PharmGKB (http://www.pharmGKB.org/). Evans transporters. Over 600 SNPs from both coding and non- reported that, in some cases, differential gene expres coding regions of transporter genes have been identi sion can predict treatment outcome; however, such fied, and the average heterozygosity (), the likelihood information appears to be treatment-specific. More that a nucleotide position will be heterozygous when recently, gene expression profiling has been used to compared across two chromosomes selected randomly determine how leukaemia cells respond to different from a population, appears to vary considerably among antileukemic agents, revealing distinct genomic re transporters. Giacomini reported segmental variation in sponses that are treatment-specific [12]. These ap amino acid diversity within transporters, with trans- proaches are now being used to identify genomic membrane domains having less variation than loop determinants of drug sensitivity in human leukaemia regions. Comparative genomics studies and functional cells. analyses in cells have confirmed that sequence variation within evolutionarily conserved regions is more likely Pharmacogenetics approaches to asthma therapy have to alter transporter function [8]. the potential to improve current treatment modalities by identifying genetic variation in drug response. In collaboration with Giacomini and Herskowitz, Este These efforts may lead to diagnostic tests for asthma ban Gonza´lez Burchard of UCSF has created the response to corticosteroids and other medications. Scott SOPHIE cohort, a group of ethnically diverse research Weiss of Brigham and Women’s Hospital presented volunteers from San Francisco who have agreed to data on two completed clinical trials investigating participate in future pharmacogenetics studies. To date, corticosteroid therapy in different populations. The Burchard and colleagues have recruited 577 individuals goal of this research is to correlate treatment response for genotyping/phenotyping studies of medications that with genotype. One of the trials was a 4-year, rando may be affected by genetic variation of membrane mized, multicentre trial of glucocorticoid therapy in transporters. Individual members of the cohort, which children. For this study, DNA samples were obtained contains approximately 125 individuals each of Mex from several hundred asthmatic children and their ican-American, African-American, Caucasian and Chi parents in the Childhood Asthma Management Program nese descent, have consented to providing DNA (CAMP). The trial has ended, but the population is still samples to be used in future pharmacogenetics studies. under investigation. The second study, an 8-week com Several UCSF investigators are currently making use of parison of flunisolide and conventional steroid therapy, the SOPHIE cohort to examine racial and ethnic was conducted in adults with severe asthma. Weiss variation in genes involved in drug response. Burchard reported that no single SNP in the corticotropin-releas noted that, although racially admixed populations make ing hormone receptor type 1 (CRHR1) gene produced up a significant proportion of the US population and it phenotypic effects in both populations, but that a is important that the research community include these common risk haplotype was present in both populations populations in genetics studies [9,10], population strati with identical effects on treatment response. The fication as a result of racial admixture can be a serious investigators modified the computer program ‘haplo. obstacle to performing genetic association studies in score’ (http://cran.r-project.org/doc/packages/haplo.score. racially admixed populations. Burchard urged the phar pdf) [13] for use as a statistical tool in data analysis. macogenetics research community to consider and, The program performs score tests for association of when necessary, correct for the presence of population traits to haplotypes in the presence of phase ambiguity. stratification in genetic association studies. In conclusion, the goal for PharmGKB, a component of William Evans of St Jude Children’s Research Hospital the PGRN, is to provide an information resource of presented data on gene expression profiling to discrimi maximum utility to the entire research community, nate molecular subtypes and drug response in child- and to stimulate future hypothesis-driven research. hood acute lymphoblastic leukaemia (ALL). Despite a PharmGKB has recently created a Web page where the greater than 99.9% similarity between host and tumour pharmacogenetics community can enter gene–drug genomes for most malignancies, several genetic charac interactions where genetic variability has been ob teristics distinguish host cells from cancer cells, the served and associated with phenotypic variability. The latter of which exhibit altered ploidy (i.e. chromosome PharmGKB Community Submission Project aims to

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 440 Pharmacogenetics 2003, Vol 13 No 7 catalogue important literature datasets in pharmacoge netics [14].

The fourth annual PGRN meeting will be held in Los Angeles, California on 8–9 March 2004.

Acknowledgement A.F.D. is under contract 263-MD-306728.

References 1 NIGMS Pharmacogenetics Research Network. http://www.nigms.nih.gov/ pharmacogenetics/ 2 Carrasquillo MM, McCallion AS, Puffenberger EG, Kashuk CS, Nouri N, Chakravarti A. Genome-wide association study and mouse model identify interaction between RET and EDNRB pathways in Hirschsprung disease. Nature Genet 2002; 32:237–244. 3 Province MA. Sequential methods of analysis for genome scans. Adv Genet 2001; 42:499–514. 4 Province MA. A single, sequential, genome-wide test to identify simulta neously all promising areas in a linkage scan. Genet Epidemiol 2000; 19:301–322. 5 Terwilliger JD, Haghighi F, Hiekkalinna TS, Goring HH. A bias-ed assessment of the use of SNPs in human complex traits. Curr Opin Genet Dev 2002; 12:726–734. 6 Schwartz S, Zhang Z, Frazer KA, Smit A, Riemer C, Bouck J, et al. PipMaker: A Web server for aligning two genomic DNA sequences. Genome Res 2000: 10:577–586. 7 Small KM, Wagoner LE, Levin AM, Kardia SL, Liggett SB. Synergistic polymorphisms of beta1- and alpha2C-adrenergic receptors and the risk of congestive heart failure. N Engl J Med 2002; 347:1135–1142. 8 Leabman MK, Huang CC, DeYoung J, Carlson EJ, Taylor TR, de la Cruz M, et al. Natural variation in human membrane transporter genes reveals evolutionary and functional constraints. Proc Natl Acad Sci USA 2003; 100: 5896-5901. 9 Risch N, Burchard E, Ziv E, Tang H. Categorization of humans in biomedical research: genes, race and disease. Genome Biol 2002; 3:2007.1–2007.12. 10 Burchard EG, Ziv E, Coyle N, Gomez SL, Tang H, Karter AJ, et al. The importance of race and ethnicity in biomedical research and clinical practice. N Engl J Med 2003; 348:1170–1175. 11 Yeoh EJ, Ross ME, Shurtleff SA, Williams WK, Patel D, Mahfouz R, et al. Classification, subtype discovery, and prediction of outcome in pediatric acute lymphoblastic leukemia by gene expression profiling. Cancer Cell 2002; 1:133–143. 12 Cheok M, Yang W, Pui CH, Downing JR, Cheng C, Naeve CW, et al. Treatment-specific changes in gene expression discriminate in vivo drug response in human leukemia cells. Nature Genet 2003; 34:85–90. 13 Schaid D, Rowland C, Tines D, Jacobson R, Poland G. Score tests for association between traits and haplotypes when linkage phase is ambiguous. Am J Hum Genet 2002; 70:425–434. 14 Altman RB, Flockhart DA, Sherry ST, Oliver DE, Rubin DL, Klein TE. Indexing pharmacogenetic knowledge on the World Wide Web. Pharma cogenetics 2003; 13:3–5.